While Nothing But Numbers has been on somewhat of a hiatus since the 2016 election, as I’ve been preoccupied with school, I am happy to announce that I expect to update the site more frequently in the coming months, especially in the upcoming week before the 2018 US midterm elections.
Similar to FiveThirtyEight and other publications, I have developed a model for both the Senate as well as the US House of Representatives for the upcoming midterm elections. While these models do not use revolutionary Facebook Likes data or demographics like my previous Presidential model, it has many notable improvements that I believe will increase the accuracy of the models. A full explanation of the methodology behind the model will be provided in a separate article published shortly, and the US House model should also be published well before the election next Tuesday.
The results and predictions of the Senate model are shown below. Keep in mind that the current composition of the Senate is 51 Republicans and 49 Democrats, and for the Democrats to take back the Senate they would need to win 51 seats in order overcome Mike Pence’s tie-breaking vote. Additionally, for clarification, I didn’t bother to run calculations on some of the non-competitive races like California, New York and Wyoming, but this does not change the overall predictions in any meaningful way.
In comparison to other models, my model is more bullish on Democrats’ chances of re-taking the Senate, as both FiveThirtyEight and the betting markets have Democrats chances of re-taking the chamber at around 15-17%, whereas I give them a roughly 1 in 4 chance. While that may seem like a small difference, 24% is roughly the same odds as world champion Mitch Moreland getting a hit in a given at bat, whereas the betting markets’ odds of 13% is roughly similar to the odds of a pitcher like Chris Sale getting a hit, something that any baseball fan would tell you is noticeable gap in hitting ability. This difference, of course, may or may not be noticeable in an individual at bat, as both players are more likely to make an out than not, but the two probabilities absolutely show different things, just as my model shows a noticeably different result than both FiveThirtyEight as well as the betting markets.
Just from a quick glance at FiveThirtyEight and the betting markets, it appears that perhaps one reason for the disparity is that my model is noticeably more optimistic than the consensus regarding the Democrats’ chances of winning senate races in Florida and Nevada, two critical states for Democrats to win in order to re-take the majority.
Still, because Republicans presently have the upper hand in enough states to give them 50 seats, while simultaneously having a solid chance of winning another four seats where Democrats are only slightly favored, the Republicans are a 3 in 4 favorite to keep control of the chamber, given the current polling. Still, a 76% chance is by no means a certainty, and Democrats have a clear path towards a majority if things break their way on election night, just as a major league hitter has a solid chance of getting a hit every time he steps to the plate.
While control of the Senate is still somewhat up in the air, it is quite clear that the composition of the Senate will fall between a specific range not too far from the current composition. Indeed, the model estimates that there is a 48% chance the Democrats will win between 48 and 50 seats, a 72% chance they win between 47 and 51 seats, an 87% chance they win between 46 and 52, and a 95% chance they win between 53 and 45, no more than a 4 seat gain in either direction. If the outcome is outside of that range on election night, there was likely either a significant error in our model or an unprecedented polling error in the generic ballot.
US Senate State-level Predictions
| State | Dem Vote% | Rep Vote% | Dem Win% | Rep Win% | Race Rating |
|---|---|---|---|---|---|
| California | 65.0% | 35.0% | 100.0% | 0.0% | Safe D |
| Connecticut | 65.0% | 35.0% | 100.0% | 0.0% | Safe D |
| Delaware | 65.0% | 35.0% | 100.0% | 0.0% | Safe D |
| Hawaii | 65.0% | 35.0% | 100.0% | 0.0% | Safe D |
| Maryland | 65.0% | 35.0% | 100.0% | 0.0% | Safe D |
| Massachusetts | 65.0% | 35.0% | 100.0% | 0.0% | Safe D |
| New Mexico | 65.0% | 35.0% | 100.0% | 0.0% | Safe D |
| New York | 65.0% | 35.0% | 100.0% | 0.0% | Safe D |
| Rhode Island | 65.0% | 35.0% | 100.0% | 0.0% | Safe D |
| Vermont | 65.0% | 35.0% | 100.0% | 0.0% | Safe D |
| Washington | 65.0% | 35.0% | 100.0% | 0.0% | Safe D |
| Minnesota | 64.2% | 35.8% | 100.0% | 0.0% | Safe D |
| Maine | 61.0% | 39.0% | 100.0% | 0.0% | Safe D |
| Michigan | 59.8% | 40.2% | 100.0% | 0.0% | Safe D |
| New Jersey | 58.3% | 41.7% | 99.7% | 0.3% | Safe D |
| Virginia | 56.8% | 43.2% | 99.3% | 0.7% | Safe D |
| Pennsylvania | 56.4% | 43.6% | 98.8% | 1.3% | Safe D |
| West Virginia | 57.6% | 42.4% | 95.6% | 4.4% | Safe D |
| Wisconsin | 55.0% | 45.0% | 95.6% | 4.4% | Safe D |
| Minnesota Special | 54.5% | 45.5% | 94.8% | 5.2% | Solid D |
| Ohio | 54.7% | 45.3% | 94.5% | 5.6% | Solid D |
| Florida | 53.3% | 46.7% | 86.4% | 13.6% | Solid D |
| Montana | 53.3% | 46.7% | 85.8% | 14.2% | Solid D |
| Indiana | 51.4% | 48.6% | 68.5% | 31.5% | Lean D |
| Arizona | 50.9% | 49.1% | 61.7% | 38.3% | Tilt D |
| Nevada | 50.8% | 49.2% | 60.7% | 39.3% | Tilt D |
| Missouri | 50.8% | 49.2% | 60.0% | 40.1% | Tilt D |
| North Dakota | 48.0% | 52.0% | 27.1% | 72.9% | Lean R |
| Tennessee | 47.8% | 52.2% | 22.7% | 77.3% | Likely R |
| Texas | 47.2% | 52.8% | 17.3% | 82.7% | Likely R |
| Mississippi Special | 44.8% | 55.2% | 4.7% | 95.3% | Safe R |
| Nebraska | 44.1% | 55.9% | 2.6% | 97.4% | Safe R |
| Mississippi | 42.3% | 57.7% | 0.6% | 99.4% | Safe R |
| Utah | 37.2% | 62.8% | 0.0% | 100.0% | Safe R |
| Wyoming | 25.0% | 75.0% | 0.0% | 100.0% | Safe R |
Overall US Senate Predictions
| Category | Democratic | Republican |
|---|---|---|
| Win Senate Majority? | 24.1% | 75.9% |
| Average Seats | 48.8 | 51.2 |
| Median Seats | 49 | 51 |
| Mode Seats | 49 | 51 |
| Max Democratic Seats | 56 | 44 |
| Min Democratic Seats | 38 | 62 |
| 538 Generic Ballot | 50.2% | 42.0% |
|---|---|---|
| Projected Generic Ballot: | 54.4% | 45.6% |
| Dems Lose Incumbent? (First Column Yes, Second Column No) | 85.4% | 14.6% |
| Reps Lose Incumbent? | 65.0% | 35.0% |
| Dems Pick Up Seats? | 40.2% | 59.8% |
| Reps Pick Up Seats? | 43.4% | 56.6% |
| Democratic Seats | Probability |
|---|---|
| 55+ | 0.2% |
| 54 | 0.9% |
| 53 | 3.3% |
| 52 | 7.4% |
| 51 | 12.3% |
| 50 | 16.2% |
| 49 | 16.8% |
| 48 | 15.1% |
| 47 | 11.4% |
| 46 | 7.7% |
| 45 | 4.6% |
| 44 | 2.3% |
| 43 | 1.0% |
| 42 | 0.4% |
| 41 | 0.2% |
| 40- | 0.1% |